Popular Mechanics (South Africa)
Mind: what you say
THEY’RE LISTENING. TO WHAT WE HAVE TO SAY. INCENDIARY OR INNOCUOUS, IT ALMOST DOESN’T SEEM TO MATTER TO THOSE WHO TRADE IN THE CURRENCY OF VOX POPULI. OPINION MINING IS THE NEW GOLD RUSH.
Opinion mining shows what we really think
HUMANS ARE SOCIAL ANIMALS. We talk, we touch, we share. In cyberspeak, we connect. As the Web develops, increasingly that’s happening on a global scale. The new wave of online interactions – the digital version of our familiar human connections – are shaping our understanding of the world in ways we never dreamt were possible.
It’s only natural that an evolution from basic communication to digital interaction would eventually progress to platforms where users can share their opinions. Here users can rate and review companies and services, or voice their views, be they social, political or of any other nature.
In this social sphere, where opinions can run rampant, negative sentiments could stimulate a chain reaction. And mass public identification: what do the terms Spur, Penny Sparrow and Hellopeter bring mind? Clearly, there’s the potential to affect profit margins, bottom lines and even have such unforeseen outcomes as protests and boycotts.
The bad news: platforms emerging thick and fast where people can voice their feedback, especially their grievances. How does a company keep track of it all?
Positive or negative, these experiences and opinions are rapidly becoming an invaluable resource to some. With opinion mining
software, corporations and institutions can pinpoint problems and respond swiftly, possibly even effectively, whether it’s about managing feedback or mitigating risk.
Finding opinions, predicting outcomes In 2016, South African opinion mining company Brandseye made headlines when it accurately predicted the outcome of the US presidential election. Using the 37,6 million Twitter conversations of 4 million users, their data matched the outcome of nine of the 11 swing states. The company also accurately predicted Brexit, Britain’s move away from the European Union. Using algorithms to process and analyse text and natural language – which includes regionalised colloquialism – data is derived that (in this case) helped predict the election outcome.
What we now call opinion mining used to go by the grand-sounding name of sentiment analysis. The older title refers to a way of looking at positive and negative feelings around topics such as brands. But opinion mining “goes a level deeper, to understand the drivers behind why people feel the way they do”, writes JP Kloppers, CEO of BrandsEye, in an article for Bizcommunity.com.
“Individual opinions are often reflective of a broader reality,” Kloppers explains. “A single customer who takes issue with a new product’s design on social media likely speaks for many others. The same goes for a member of the public who takes to a political campaigner’s web page to praise or criticise the policies proposed.”
In essence, opinion mining looks at every- thing we say online and compiles massive amounts of data around our emotions. It is also capable of grouping individuals with shared ideas and opinions. Sentiment analysis, on the other hand, is an advanced form of social media listening that systematically identifies, extracts and studies media for polarising opinions with positive or negative sentiment.
Sentiment analysis versus opinion mining In a Researchgate study that looks specifically at Tweets referencing automotive manufacturers, researchers used sentiment analysis to link polarity and emotions to luxury brands.
Using a dataset of 730 000 Tweets, the researchers found that BMW performed better in the “joy” category than MercedesBenz and Audi. The last two manufacturers shared a higher percentage in the “sadness” category than BMW. The study’s abstract notes that BMW’S Tweet percentage was 72 per cent positive, compared with Mercedes-benz (79 per cent) and Audi (83 per cent).
When it comes to intricate detail, opinion mining betters sentiment analysis. “A fast food chain might be interested to know that relative to their closest competitor, many consider their portion size too small, though their friendly customer service is a plus. Yet it’s possible to delve deeper still. The topic of customer service could be further broken down into the subcategories of turnaround time, order correctness and delivery time. The business may have great in-store turnaround, but fail when it comes to deliveries. Knowing which issue to target – and why – is key,”
writes Kloppers.
Mitigating risk On Sunday, 9 April 2017, the forceful removal of a passenger from a United Airlines flight was captured on video. Viewers see a passenger, with bloodied face, being dragged down the aisle of a plane. Media worldwide picked up on the story like wildfire. The episode opened the airline up to censure from millions of Internet users.
The brand’s name was mentioned more than 762 000 times on Facebook, Instagram and Twitter the day after it went public (10 April), according to media monitoring site Brandwatch, as reported by news website The Drum. Further showing the depth of damage caused by the events of 9 April, the news site said: “This was in comparison to the 135 000 mentions after earlier #legginsgate incident late last month.”
This number preceded the more than 290 million views when the video trended in China on local microblogging platform Weibo. It drew outrage from Chinese nationals because the passenger was of Chinese ethnicity.
United Airlines looked bad as millions of social media users slammed the way the situation was handled and how the airline responded. In the days afterwards, United stock dropped 1,1 per cent, costing the airline more than $225 million in market capitalisation.
In cases like this, data derived from opinion mining could prove an invaluable tool to mitigate risk. When analysing blogs, Facebook posts, Tweets and (in some cases) email and instant messaging conversations, institutions can focus on not only negative sentiment, but the source or cause of the sentiment. The data could serve as a preventive measure or guideline to remedy the source of the problem.
It’s not just a PR exercise Opinion mining could focus on specific regions areas, by looking at location tags. This has the potential to aid business best practice, or service delivery for government departments. The data could be targeted to problem areas; for example, roads where trees often fall over, flooding often occurs, or that carries heavier traffic than before.
With a eye on the public and a firm hand on reputation, opinion mining could be a great solution to limit problems before they even arise. Maybe that scathing Tweet wasn’t, after all, such a bad idea. PM
“Individual opinions are often reflective of a broader reality”